🧠 bertopic-admissions-mmr-keybert

This model is a fine-tuned BERTopic model for clustering university admissions-related questions and documents using Maximal Marginal Relevance (MMR) and KeyBERT-based keyword generation.

🏗️ Model Details

Base Model: BERTopic (HuggingFace Transformers + UMAP + HDBSCAN)
Embedding Model: all-MiniLM-L6-v2
Keyword Method: MMR + KeyBERT
Training Data: 50-question CSV dataset on university admissions topics
Date Trained: April 2025

📊 Intended Use

• Question clustering for FAQ and chatbot systems
• Identifying user intent for university-related inquiries

🧯 Limitations

• Small training dataset (50 rows)
• English-only
• May group distinct topics if vocabulary overlaps

📁 How to Use

from bertopic import BERTopic

# Load model
topic_model = BERTopic.load("your-local-folder-or-hf-repo-name")

# Transform new docs
topics, probs = topic_model.transform(docs)